speed up submit-grid-jobs a lot by only calling condor_q once

1 files changed, 19 insertions, 8 deletions

diff --git a/utils/submit-grid-jobs b/utils/submit-grid-jobs
index de022e1..8f54c1e 100755
--- a/utils/submit-grid-jobs
+++ b/utils/submit-grid-jobs
@@ -394,7 +394,7 @@ def release_job(submit_file):
 
     return -1
 
-def get_job_status(submit_file):
+def get_job_status(submit_file, data=None):
     """
     Check to see if a given grid job is finished. Returns the following statuses:
 
@@ -412,16 +412,20 @@ def get_job_status(submit_file):
     http://pages.cs.wisc.edu/~adesmet/status.html.
     """
     log.debug("condor_q -json")
-    output = check_output(["condor_q","-json","--attributes","SubmitFile,JobStatus"])
 
     head, tail = split(submit_file)
 
-    if output:
-        data = json.loads(output)
+    if data is None:
+        output = check_output(["condor_q","-json","--attributes","SubmitFile,JobStatus","--constraint",'SubmitFile == "%s"' % tail])
 
-        for entry in data:
-            if entry['SubmitFile'] == tail:
-                return entry['JobStatus']
+        if output:
+            data = json.loads(output)
+        else:
+            data = []
+
+    for entry in data:
+        if entry['SubmitFile'] == tail:
+            return entry['JobStatus']
 
     # If there's no entry from condor_q the job is done. Now, we check to see
     # if it completed successfully. Note: Jobs often don't complete
@@ -482,6 +486,13 @@ def main(conn, dqxx_dir, max_retries, max_jobs):
 
     stats = {}
 
+    output = check_output(["condor_q","-json","--attributes","SubmitFile,JobStatus"])
+
+    if output:
+        data = json.loads(output)
+    else:
+        data = []
+
     for id, filename, run, uuid, gtid, particle_id, max_time, state, nretry, submit_file in results.fetchall():
         if state not in stats:
             stats[state] = 1
@@ -505,7 +516,7 @@ def main(conn, dqxx_dir, max_retries, max_jobs):
                 c.execute("UPDATE state SET state = 'RUNNING', submit_file = ?, nretry = ? WHERE id = ?", (abspath(submit_file),0,id))
         elif state == 'RUNNING':
             # check to see if it's completed
-            job_status = get_job_status(submit_file)
+            job_status = get_job_status(submit_file, data=data)
 
             if job_status in (0,1,2,4):
                 # nothing to do!